Re: Debian Vision (version 1.0)

[Peter Tobias <tobias@et-inf.fho-emden.de>]

On Mar 25, Jim Pick wrote


IMHO the posting from Jim is VERY important. We need to set some goals.
It seems that the Debian developers don't have the same goals for the
Debian distribution. We need to specify which goals are important for us.

We should take Jims mail as base for a list of things we want to
achieve with Debian. Maybe this list should also be available on
the web server (to show the Debian user what we want to achieve
with Debian).


> What's the Debian project all about?
> ------------------------------------
> 
> 1.  Community - active mailing lists, maintainers who answer your e-mail, 
>       extremely smart people, cool spam filter, really interesting 
>       discussions and flamewars...
> 
> 2.  Good software - Linux, GNU, Perl, Apache, GIMP, etc., etc., etc.
> 
> 3.  Non-commercial - nobody's trying to make a buck off of me

We shouldn't "sell" the Debian distribution (the distribution has to be
free for all users). We should suggest that people that are distributing
Debian CD should donate some money (leave it up to the people how much
money they want to give) to the Debian project (it shouldn't be a
requirement!). Just tell them that with a (small) donation they make
sure that this is a way to make sure that the Debian project will survive
and can pay their bills.

> 4.  dpkg - overall, it's pretty slick, and addresses a lot of problems that
>       haven't been addressed in the commercial software world
>           - database of all files in system
>           - install/removal scripts
>           - source diffs
>           - access methods (ftp, nfs, etc....)
>           - not monolithic, so it will adapt quickly
> 
> 5.  It actually invites it's users to actively participate in the project
>
> 6.  Large numbers of maintainers - so each can focus on a small aspect of
>       the system - which improves the quality

The more maintainer we have the more problems we will have with managing
the whole distribution (more places where an installation could fail). If
we would allow more maintainers we have to make sure that only packages
that won't cause trouble during the installation process will be moved
to the stable tree.

> 7.  Documentation - we have potential to be the best documented OS on the 
> planet
>
> 8.  Cross-platform - potentially
> 
> 9.  Support - anybody posting a simple question to debian-user usually gets two
>       or three solutions to their problem within an hour or so
> 
> 10. Respectability - the word-of-mouth on Debian is very good - how do you 
> think
>       we got so many users?

The Debian distribution also has the reputation of being hard to
install (compared to Redhat, S.u.S.E and FreeBSD) and to sometimes
contain fatal errors (libdl problem) that prevent an installation
at all (unless the user knows how to fix it himself).

> 11. The bug system - it does the job admirably
> 
> 12. Security - we seem to be pretty attentive towards it

I would reduce it to the following topics:

   - a freely available distribution
   - it _should_ use only free software (with the option to use non-free
     stuff if it is necessary) for the main distribution tree
   - stability
   - security
   - good support (bugfixes)
   - documentation
   - multiplatform

The other topics are mostly ways to achieve the goals mentioned above.

> Who's our community?
> --------------------
> 
> 1.  Unix network administrators (particularily academic)
> 
> 2.  Linux hackers
> 
> 3.  People running ISP's
> 
> 4.  Web application developers
> 
> 5.  College students
> 
> 6.  Business types who are building networks that need to be 
>       remotely adminstrated, and where cost per node is a factor
> 
> 7.  MIS types - Linux works quite well as a file/print server
>       and also as a relational database server
> 
> 8.  HAM radio types - eh, Bruce?
> 
> 9.  Electrical engineers - like me
> 
> 10.  Hardcore computer hobbyists - installing Linux as a second
>     (or third) OS on their home computers
> 
> 11. Computer scientists - all those Lisp programs...
> 
> 12. Professional software developers - CVS, gcc, emacs, java, 
>       Perl, Tcl/Tk, Python,etc...
> 
> 13. Internet junkies - exmh, Gnus, IRC, Apache, Sendmail, etc..
> 
> 14. Computer graphics fans - POV-ray, GIMP, xv, imagemagick, etc...


A more general description for the groups mentioned above is:

                         -------------------------------------------------
The optional             |                 Debian User                   |
"Workstation" stuff      |                    System                     |
                         -------------------------------------------------
The main purpose:        |    Server     |  Development  |     Other     |
                         -------------------------------------------------
The common base:         |              Debian Core System               |
                         -------------------------------------------------
 
All use the core system. It is very important to distribute an error
free core system (it affects all users).


> What could be done better?
> --------------------------
> 
> 1.  Testing
> 
> 2.  Documentation (where I'm concentrating my efforts)

We need more documentation of "debian specials" ...

> 3.  Fewer flamewars and ultimatums - I really hate these "I'm quitting" posts
> 
> 4.  dpkg user interface
> 
> 5.  Organization (notice that I rank this as #5, not #1)
> 
> 6.  On-line distribution (improved dselect access methods, load-balancing,
>       more control over mirror sites, etc.)
> 
> 7.  Off-line distribution - Red Hat is addressing the shrink-wrapped software
>       market - nobody's doing the equivalent with Debian
> 
> 8.  More "openness" - I don't like the idea of having a "debian-private".  Most
>       users I've spoken to don't like it either.

NO. Debian-private shouldn't be available to non-developers. It could
be bad PR for us. Family problems should stay in the family.

> 9.  Maintainer verification - we need to address a valid criticism
> 
> 10. Tight schedules - it's hard as a developer to do everything inside the 
> allotted
>       timeframe.  Schedules tend to slip -- and not enough time is allocated 
> towards
>       doing testing.
> 
> 11. Constant infrastructure changes - the mailing lists have be changed many, 
> many
>       times over the last year or so.  Maybe this is somewhat unavoidable - 
> but it
>       would be nice to have a more stable environment.
> 
> 12. Too many single points of failure - people go on vacation, and things get 
>        delayed.

Maybe we need some kind of "backup developers" who don't maintain packages
(or very few) and who will fix bugs that need to be fixed if the normal
maintainer can't do it in time.

> 13. The unstable distribution is too unstable at times to do development on -- 
> it
>        can be a big productivity drain trying to figure out why other people's
>        packages screwed up your system

14. We need a way to separate the program bugs from the packaging bugs.
    IMHO we need two different kind of bugreports:

    - bugs that affect the installation. Maintainers should fix this within 2 weeks.
      If this is not possible within the two weeks we should have a few people who
      will fix the bugs and submit the patches back to the maintainer. These kind
      of bugs have all to be fixed for a new Debian release (you'll get a nice
      list of things to do before a release then).
    - the other bugs (Probably a lot of them could be forwarded upstream)

    How can we achieve this? One solution would be to add a new
    optional field to the bug reports and to have someone to dicide
    about the bug reports that don't have such a field.

    Having this two kind of bugs allows us to provide a distribution
    that other people can install without problems. I often read
    "hey, Debian is quite nice, but it is hard to install and often
    have bugs that cause you trouble during the installation". That
    is really bad PR for Debian.


Thanks,

Peter

-- 
 Peter Tobias                                EMail:
 Fachhochschule Ostfriesland                 tobias@et-inf.fho-emden.de
 Fachbereich Elektrotechnik und Informatik   tobias@debian.org
 Constantiaplatz 4, 26723 Emden, Germany     tobias@linux.de